[Debian-med-packaging] dcmtk_3.7.1+dfsg-1~0exp0_amd64.changes REJECTED
Étienne Mollier
emollier at debian.org
Sun Oct 18 10:24:14 BST 2026
Hi Andrew,
I add Mathieu in the loop for awareness.
Andrew McMillan, on 2026-10-17:
> The LLM Review has found some evidence of files that should
> not be in the package, either because they are licensed in a
> way that is not distributable, or the compiled files are in
> place when the sources have been excluded. I looked in the
> package myself, and while some of this seems true, there's a
> different read on what is written there which says that the
> LLM is just misunderstanding what is going on.
>
> Can you reply with your understanding of these issues (see
> full review at link below)?
> https://dfsg-new-queue.debian.org/reviews/dcmtk
Items revolving around Unicode files recently surfaced with the
following Lintian error:
E: dcmtk source: license-problem-old-unicode [debian/copyright]
Patterns identified by your inference engine seem therefore
somewhat accurate to me. While I attempted removal of affected
files a first time yesterday, binary artifacts present in
upstream source code went in a blind spot during my review.
Apparently, I also missed a couple of items that are still
present in my first repack: they weren't flagged by lintian and
due to the old Unicode notice being quite unlike regular license
terms, I also missed them when looking that up by hand.
When I examined #854209 (pointed to by lintian-explain-tags),
I have been under the impression that other projects managed to
obtain newer copies of the affected files with the newer
terms[1]. This is not something I managed to get somewhere
with though, so I am now investigating whether entire removal of
the oficonv, in favor of using iconv as provided by the glibc,
is manageable. If that works, then I should be able to follow
up with another upload with extended files exclusion.
[1]: https://www.unicode.org/copyright.html
Mathieu, I assume dcmtk provides oficonv for good reasons, so
if you believe there are hard blockers to using plain iconv
instead, please don't hesitate to raise your concerns.
Ultimately, this problem sounds like something that has to be
brought upstream.
In hope this clarifies things,
--
.''`. Étienne Mollier <emollier at debian.org>
: :' : pgp: 8f91 b227 c7d6 f2b1 948c 8236 793c f67e 8f0d 11da
`. `' sent from /dev/pts/1, please excuse my verbosity
`-
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-med-packaging/attachments/20261018/dff9a5c6/attachment-0001.sig>
More information about the Debian-med-packaging
mailing list